Why smarter AI models could drive up compute prices 10x
Summary
Dwarkesh analyzes the divergence between AI labs' 10x revenue growth and only 3x compute scaling, attributing it to rising margins and compute prices. He argues that as AI becomes more valuable, compute scarcity will favor efficient models, potentially pricing out current applications. The episode concludes that supply constraints will persist until automation drastically lowers chip production costs.
Topics discussed
Anthropic's revenue growth vs compute scaling trends
Three factors bridging the revenue-compute gap
Rising compute prices and Google/Anthropic case study
Monetizing smarter models and labor economics
Market consolidation and the Alchian-Arrow effect
Compute scarcity limits and hardware bottlenecks
Sponsor: Mercury Command AI for bookkeeping
Future compute costs and economies of scale
Listen ad-free on Castria